Google Threat Intelligence Group has identified fully autonomous AI cyberattacks: systems that independently analyze updates, find vulnerabilities, create exploits, and launch attacks; analysts also discovered a platform managing more than 100,000 accounts and cases where cloud resource breaches resulted in costs up to $1m.
AI Executes Entire Hacking Cycle Without Human Involvement
The report describes a system capable of automatically analyzing software updates, detecting vulnerabilities, generating exploits, and launching attacks without manual intervention.
North Korean Attacks on Crypto Firms via Fake Recruiting Sites
Analysts recorded attackers impersonating recruiters, directing employees to counterfeit websites, and modifying Claude Code settings so that a backdoor would install automatically upon launch.
Large-Scale Botnets and Exploitation of Cloud Resources
A platform was discovered managing over 100,000 social media accounts simulating human behavior to spread political content; separately, cases of corporate cloud breaches were noted, causing financial losses up to $1m for affected companies.
